http://hdl.handle.net/11422/86 Sun, 19 Jul 2026 17:52:27 GMT 2026-07-19T17:52:27Z http://hdl.handle.net/11422/29686 Title: Sobre a construção da curva carga-recalque equivalente de prova de carga convencional a partir de prova por método bidirecional Author(s)/Inventor(s): Lima, Leandro Soares Advisor: Lopes, Francisco de Rezende Abstract: This work addresses the behavior of piles in bidirectional load tests, compared with conventional tests, making use of the Finite Element Method. The SigmaW software is used, in an axisymmetric solution, with linear elastic, hyperbolic and elasto-plastic models of soil behavior. The study seeks to evaluate the differences in shaft friction mobilization in the two loading modes. Regions with displacements in opposite directions (ascending and descending) and the differences in stress distribution in the soil are observed. Firstly, a parametric study (hypothetical cases) evaluates different length/diameter ratios and ground profiles. Initially, a simple, elastic-plastic approach is used, with the prediction of settlement for the service load by the Randolph method and the failure load by a theoretical method of calculation of load capacity. Then, the connection between the two states (a point on the load-settlement curve in service and a vertical asymptote at the failure load) is made with the Van der Veen curve. This work also presents a case study of two nearby auger piles, one subjected to a bidirectional load test and the other to a conventional test. The Mohr-Coulomb elasto-plastic soil model was used in numerical modeling, with soil resistance and deformability parameters estimated from a SPT and a sensitivity study. The results of the analyzes allows some suggestions for extrapolating the results of bidirectional tests to the behavior of the piles when loaded from the top – actual working condition –. Publisher: Universidade Federal do Rio de Janeiro Type: Dissertação Sat, 01 Jun 2024 00:00:00 GMT http://hdl.handle.net/11422/29686 2024-06-01T00:00:00Z http://hdl.handle.net/11422/29684 Title: Interação entre modos de flambagem de perfis de aço formados a frio: análise e método de dimensionamento estrutural Author(s)/Inventor(s): Matsubara, Gustavo Yoshio Advisor: Batista, Eduardo de Miranda Abstract: Cold-Formed Steel (CFS) refers to structural members formed by bending flat, thin plates at room temperature into shapes that enhance their load-carrying capacity compared to the original plate, achieving a favorable strength-to-weight ratio. While these structures offer significant advantages, it's crucial to consider their elevated slenderness values. This can lead to the formation of thin-walled structures, which are prone to buckling modes, including local (L), distortional (D), and global (G) buckling. Global buckling includes bending (GF) and flexural-torsional (GFT) modes. These buckling modes can interact, resulting in local-global (LG), local-distortional (LD), distortional-global (DG), and local-distortional-global (LDG) interaction modes. Such interactions can significantly reduce the resistance of these structures compared to isolated modes. The primary method for designing cold-formed steel under axial compression is the Direct Strength Method (DSM). The DSM considers design methodologies for CFS members subjected only to local (L), distortional (D), and global (G) buckling modes, as well as the interaction between local-global (LG) buckling modes. However, the DSM does not cover the interaction between other buckling modes. This study proposes modifications to the DSM to address these interactions. The suggested modifications have been shown to effectively estimate the resistance of bars under axial compression, supported by both numerical and experimental results. Publisher: Universidade Federal do Rio de Janeiro Type: Tese Fri, 01 Mar 2024 00:00:00 GMT http://hdl.handle.net/11422/29684 2024-03-01T00:00:00Z http://hdl.handle.net/11422/29683 Title: Previsão de séries temporais de esforços em risers flexíveis durante a fase de instalação por meio de redes neurais convolucionais Author(s)/Inventor(s): Barbosa, Felliphe Góes Fernandes Advisor: Sagrilo, Luis Volnei Sudati Abstract: Offshore projects for flexible pipe installation play a relevant role in connecting wells to platforms. The development of these projects involves numerical simulations to analyze the behavior of equipments, with a particular focus on flexible pipes. Often, environmental loadings generated by ocean waves are modeled deterministically, known as regular wave modeling, due to its computational efficiency. However, this approach is a substantial simplification compared to the actual motion of waves, which is better represented by modeling with irregular waves. This work proposes a methodology to make the use of irregular wave modeling feasible in installation analyses, employing artificial neural networks. Specifically, convolutional neural networks are trained with data generated from short length finite element-based numerical simulations. The developed model focuses on modeling the tension at the top of the pipe and the touchdown zone for various configurations that occur during the installation of a laze wave flexible pipe. Additionally, the model is adaptable to different environmental conditions, providing a substantial reduction in simulation time. Publisher: Universidade Federal do Rio de Janeiro Type: Dissertação Fri, 01 Dec 2023 00:00:00 GMT http://hdl.handle.net/11422/29683 2023-12-01T00:00:00Z http://hdl.handle.net/11422/29666 Title: Discurso de ódio em redes sociais: uma abordagem automatizada para identificação em conteúdo em português Author(s)/Inventor(s): Oliveira, Felipe Ramos de Advisor: Ebecken, Nelson Francisco Favilla Abstract: Social media platforms play a pivotal role as means of human interaction in the contemporary era, providing platforms for communication and expression. However, the increasing prevalence of hate speech on these platforms poses significant risks to individuals and communities. Detecting and addressing hate speech is particularly challenging in languages like Portuguese due not only to its rich vocabulary, complex grammar, and regional variations but also to the scarcity of annotated datasets for this purpose. Based on this, this thesis presents the largest annotated dataset in Portuguese for automatic hate speech detection. The creation of this corpus is accompanied by an extensive literature review containing different definitions of hate speech, data availability, analyses of the peculiarities of hostile content propagation online in Brazil, and language model architectures. To assess the viability of the dataset, models with architectures representing the state of the art in the literature were used, with tests evaluating the use of BERT and GPT-2. Publisher: Universidade Federal do Rio de Janeiro Type: Tese Mon, 01 Apr 2024 00:00:00 GMT http://hdl.handle.net/11422/29666 2024-04-01T00:00:00Z